Gyaata
The essay
Almost everything written about the forgetting curve online quotes a percentage from it. Those percentages come from one man memorising invented syllables in the 1880s, and they were not measuring what people think they were measuring. Here is what the experiment actually did, and what it does and does not license you to conclude about your own revision.
1885
Hermann Ebbinghaus published Über das Gedächtnis in 1885, reporting experiments he had run on himself in 1879–80 and again in 1883–84. He was the only participant. His material was lists of invented nonsense syllables, chosen precisely because they carried no meaning — he wanted memory stripped of everything you already know, which is the opposite of the situation you are in with a Polity textbook.
And the thing he measured was not recall. It was savings in relearning: he learned a list, waited, then learned it again, and recorded how much less time the second pass took. A saving of some fraction is not the same statement as “you remember that fraction”, and the two come apart. Every popular retelling of this curve quietly converts one into the other.
The shape, and only the shape. There are no numbers on either axis here on purpose — see below for why the ones you usually see printed on this curve are not numbers about you.
What survives scrutiny
Murre and Dros reproduced the curve in 2015, running the same method of savings on a single subject, and got results close to the original. So the shape is real and it is not folklore: forgetting is steepest immediately after you learn something and flattens out afterwards. That is a genuine finding and it is the reason reviewing on a schedule beats reviewing when you happen to feel like it.
What does not survive the trip is the numbers. A figure derived from one person relearning meaningless syllables is not a measurement of how much of a chapter you will have lost by Sunday. Your material has meaning, structure, and hooks into things you already know — all of which change how it decays. Anyone printing a specific percentage next to the words “you forget” is quoting an experiment that did not ask your question.
Not a softened one either. If we cannot source a number, we do not print a vaguer version of it and call that honest — we leave it out and tell you why. The two things this page is willing to assert are the shape of the curve and the method that produced it, and both are cited at the bottom.
The part that matters to you
The practical consequence of a steep-then-flattening curve is that review timing matters more than review volume: a page reread the same evening buys you much less than the same reading spread across the weeks before you need it. That much follows from the shape.
The part almost every study tool leaves out is that “before you need it” is a real date. How far apart your reviews should sit depends on how long you have to hold the material, so a scheduler that does not know when you sit the paper is guessing at the one input that decides the answer. That is the whole of what Gyaata does differently: you give it the date, and it plans the reviews inside it.
Sources
Ebbinghaus, H. (1885). Über das Gedächtnis: Untersuchungen zur experimentellen Psychologie.
The original. One participant — Ebbinghaus himself. The data were collected in 1879–80 and again in 1883–84; 1885 is the publication year, not the year of the experiments.
Murre, J. M. J., & Dros, J. (2015). Replication and Analysis of Ebbinghaus’ Forgetting Curve. PLoS ONE 10(7): e0120644.
A successful replication, again by the method of savings, with a single subject spending roughly 70 hours learning and relearning lists. It also examines how savings relates to the percentage-correct measure people assume the curve reports.
That is the complete list, because those are the only two claims on this page that need one.